Snowflake-to-Fabric Iceberg Mirroring: Scheduling Limits and Decimal Type Bugs Documented

·1 views

Engineers integrating Snowflake Iceberg tables into Microsoft Fabric via mirroring have documented two significant issues uncovered through support investigations. First, Fabric currently offers no configurable mirroring schedules or replication windows, and stopping then restarting the service triggers a full data reload rather than resuming from the last change-data-capture position. Second, a type mismatch bug causes Spark reads to fail when a Parquet file stores a decimal column as INT32, even though the SQL analytics endpoint handles the same data without error. The team traced the discrepancy across Snowflake, Fabric table schema, and the underlying Parquet files, and submitted findings to Microsoft support. As a workaround, non-vectorized Spark reads resolved the decimal issue, though with notable performance and capacity trade-offs; the team is next evaluating whether routing Snowflake replication from AWS to Azure improves overall economics.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

AT&T Cuts AI Costs 80% by Shifting Routine Tasks to Open-Source Models

AT&T has significantly reduced its AI spending by migrating high-volume, routine tasks like call transcription and customer service to open-weight models, with its chief data and AI officer Andy Markus reporting costs down as much as 80 percent compared to earlier this year. Open models now account for 40 percent of AT&T's AI usage, up from 20 percent in May, with Markus projecting that figure could reach 60 percent in the coming months. The company is specifically using Meta's Llama and Google's Gemma for customized tasks, while deliberately avoiding Chinese open models such as DeepSeek and Moonshot despite their lower costs. Closed frontier models from OpenAI and Anthropic are being retained for more complex tasks like heavy coding and multimodal work where open alternatives fall short. The broader trend is echoed by companies like Airbnb and Deloitte, reflecting a growing corporate strategy of reserving expensive frontier AI for specialized needs while offloading simpler, high-volume workloads to cheaper open-source alternatives.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

AI Text Overload Sparks Shift Toward Visual Answers in Chat Interfaces

As AI-generated text becomes effortless and abundant, its perceived value has sharply declined because readers rarely engage with long AI-produced documents. The core problem, according to a developer writing on DEV Community, is a format mismatch: text buries answers while readers want them surfaced instantly. The author argues that for questions involving numbers, status updates, and comparisons, visual formats like charts and dashboards are far more effective than prose. To address this, the developer built mcp-dashboards, a free open-source MCP server that renders interactive charts and dashboards directly inside AI chat conversations. The tool, compatible with Claude desktop, leverages the MCP standard updated in January 2026 and processes data locally without sharing it with third-party services.

0 comments Read more at DEV Community

ProgrammingHacker News ·

OpenJDK Proposes JEP 544 to Bring Ahead-of-Time Code Compilation to Java

OpenJDK has published JEP 544, a proposal aimed at introducing ahead-of-time (AOT) code compilation to the Java platform. AOT compilation allows Java code to be compiled into native machine code before runtime, potentially improving startup performance and reducing warm-up time. The proposal is part of ongoing efforts within the OpenJDK community to enhance Java's execution efficiency. Details of the proposal are available on the official OpenJDK website for review and community feedback.

0 comments Read more at Hacker News

ProgrammingDEV Community ·

EU Eyes Age Verification and Social Media Curbs to Protect Children Online

European Commission President Ursula von der Leyen outlined a stronger EU policy push for child online safety at the European Summit on AI and Children in Copenhagen on May 12, 2026. Her remarks emphasized platform design accountability, age verification, and possible social media access restrictions for minors. While no new legal obligations were introduced, the speech signals where the Commission is headed following the existing Digital Services Act and Digital Markets Act. An open-source age-verification tool is being considered for EU-wide deployment, with a demonstration rollout in Denmark planned for summer. A new expert panel on child safety online was also announced to advise on these evolving measures.

0 comments Read more at DEV Community